The Agent Skills Directory

[SAFE]: The skill consists of architectural documentation and technical integration examples for the inference.sh platform SDK. No malicious code, hidden instructions, or suspicious behaviors were detected.\n- [EXTERNAL_DOWNLOADS]: The documentation provides standard instructions for installing the official @inferencesh/sdk package from the public npm registry and refers to vendor-owned resources on the inference.sh domain.\n- [COMMAND_EXECUTION]: The documentation includes illustrative code snippets for developers, such as implementing a calculator tool handler using eval() or reading local files via the Node.js fs module. These examples are template code provided for end-user implementation and are not executed by the skill itself.\n- [PROMPT_INJECTION]: As an SDK for building AI agents, the documentation describes patterns for processing external data (messages, file uploads), which represents an inherent surface for indirect prompt injection if the resulting implementations lack proper sanitization.\n
Ingestion points: User message inputs and file uploads (documented in SKILL.md and references/files.md).\n
Boundary markers: Not explicitly demonstrated in the high-level usage examples.\n
Capability inventory: SDK-built agents can be configured with capabilities like web search, code execution, and external webhooks (references/tool-builder.md).\n
Sanitization: Standard validation and sanitization are left to the implementation of the tool handlers by the developer.

javascript-sdk